Preparing for Machine Vision
| By: Winn Hardin, Contributing Editor
From line tests to targeted technical manuals, vision end users share their thoughts on how to prepare for installing a new system.
Vision engineers at major US manufacturing companies say they try out new vision systems long before an installment request comes from their people. In this way, the system is already qualified before a new product line needs immediate assistance.
Unlike a standard PLC, machine vision systems require careful consideration and planning before installing a final solution. We've asked several in-house engineering groups for large manufacturers what steps they take before placing a system. Bearing the responsibility of both the machine vision manufacturer and integrator, these engineers know that they have the highest level of responsibility - like the neighbor who does work on his friend's house, he can never walk away. At the same time, however, these engineers must deal with the same reality that all vision companies face - you can't be there all the time and still do your job.
In walking this line between support and reality, in-house engineers tell Machine Vision Online how they prepare for the coming of automated processing. By looking at their preparation roadmap, vision companies can better serve their customers and gain repeat business.
In house test beds
Large companies with the financing and wherewithal to have in-house vision engineers don't wait for an application to come along before shopping for a new system. "We only install off the shelf machine vision systems," said Steve Spicer, vision systems engineer at 3M. "What we do is purchase a vision system, research for a while - about 1 to 5 years - and if there's an application that it'll work on, we install it and then buy another. Sometimes you have more than one choice, and for others there's just one choice that's a logical fit."
Testing and evaluation are also a big part of Corning Inc.'s standard operating procedure. Scott Deming, manager of Corning's corporate engineering group, focuses mainly on microassembly in the photonics and advanced life sciences division, in addition to serving the entire corporation where vision is concerned. "We've been evaluating vision since the early '80s when the industry was in its infancy. There were roads that we started down that turned out to be dead ends. The main thing driving us right now is the tight [manufacturing] tolerances we're trying to achieve and getting some of the tedious mundane tasks away from our operators."
During the past couple of years, Deming has set up a test bed that contains sample applications of the various manufacturing methods used at Corning. "One typical approach we use is that we work with the manufacturing facility and understand where their needs are from an automation and vision perspective. Once we've identified the needs, we'll pick out a particular application and bring it into the lab to demonstrate the feasibility of a vision system." If the facility engineers and operators like what they see, Deming says "then we get into the financial aspects of what could be gained by the use of vision and automation. Then we prototype the system in the manufacturing environment."
Working with the line supervisor or facility manager is also critical to Spicer's operation. "If a product exists for a particular application, we bring it into the facility and try them on the line so that the plant is making the decision…We never get a 100 percent solution because there are just some things that the system can't differentiate or that it just can't find. We give the plant the option, saying, 'we can do these things, but to find that last defect we might need a completely different lighting arrangement. And since that's only about 1 percent of your rejects, we tell them to make the final decision on whether it's financial feasible to do that," he said.
Both Spicer and Deming tend to keep to the larger, more established machine vision suppliers. "The larger names are more likely to have point and click programming. If someone [smaller] comes in here and shows me that their system can do it, we'll look at it but we're still interested in long term support and that's a big issue that goes into our thought process about what we're going to install," Spicer said.
"After the system is implemented and installed … there has to be some level of support available to the manufacturing facility. These days, the products are changing quicker than they ever have, so we're constantly dealing with how do we train and teach the vision system a new feature or product," Deming said. "Partnering with the right supplier is important…and is likely the key to success of the project. When you're done with installation, you've got a significant investment. When [the supplier] changes parts for the system, a business can be left with a system that no longer functions. So we look for forward-looking companies that constantly try to improve their software and hardware - [and still] … support the legacy equipment."
This is not to say that Deming only selects large machine vision companies. "We've gone with a four-man company in Upstate New York that developed a system that works great on one application. But when he closes his doors, it leaves you without any support unless you want to pay a large consultant retainer, that is," he said.
While neither man could stress enough the significance of support to a successful implementation, system flexibility and ease of use are also crucial in today's workplace. "Flexibility is the name of the game these days. With some of our new businesses, like photonics, we may have 20 to 30 new product introductions on one assembly line and that requires a lot of flexibility to deal with," Deming said. "Also the system needs to be user friendly, because people don't stay at jobs as long as they used to. We constantly have movement within our workforce, so the system needs to be fairly simple to use," Deming said.
3M also realizes that much of the system set up and trouble shooting will have to be conducted by plant maintenance personnel, while training a new product may bring the attention of a process engineer. "If a product line changes, we can help but we make sure they're capable of handling it. Training personnel is great, but you may spend a lot of time on training and then the maintenance or process engineer only touches the system once every six months. It's not a good value unless you have a lot of these systems and you know that you're going to be dealing with them regularly. We provide documentation through a contract technical writer that does documentation for the specific system. We do train, but not on the entire system - only the tools and functions that the application is using. General training is too diverse; we just focus on the need to know," Spicer said.
Preparation for small companies
Obviously, the 3Ms and Cornings make up a small percentage of the manufacturing market. Not every company is going to have the advantages of a full-time vision engineering staff, in-house test beds or even the skills of a full-time technical writer to tailor documentation to a specific application line. For these companies, Deming offers the benefits of 25 years of experience.
"Decision makers need some expertise to drive the process. Obviously, there's some level of knowledge that you need internally even if you're working with outside integrators - just to make sure that somebody's not trying to sell you something that doesn't have a chance of working," Deming said.
"One of the ways internally that we keep up to date on the technology is that we'll attend vision shows and conferences. …Those shows allow you to network with other people in the industry and understand the technologies that are available…It's critical to know the system before making a decision," he added. Deming has made his own mistakes. In the early '80s, Corning tried to make an automated loading system for catalytic converters. "We had a robotic partner and we found that after a year and half we were trying to push the technology too hard…As a result of that project, people were very skeptical of using vision because of the early failures."